Millions of Americans have welcomed artificial intelligence into their homes in recent years via virtual assistants like Alexa, Bixby, Cortana, and Siri. Privacy concerns aside, they’ve revolutionized the way we interact with our homes and plan our lives. But would those same Americans feel comfortable entrusting AI with their home insurance?
Your answer to that question might not matter much — because it’s already happening whether you want it to or not.
“AI is having a profound impact on the insurance industry across the board, from strategy and policy pricing to marketing and administration,” the CEO of AMPLYFI, Chris Ganje, told Forbes earlier this year.
In 2019, the primary way AI is changing home insurance is during the shopping process. Ten years ago, getting home insurance quotes could be an exhausting process. You’d have to talk to an agent, provide hundreds of details about your home and your family, and sometimes even schedule a home visit with an inspector. To compare prices between different insurance companies, you’d have to repeat that same process over and over again. The digitization of insurance shopping over the past decade has made it much easier to compare quotes from different providers, but AI has made it possible to get lightning-fast quotes.
Quotes in 15 minutes or less
Today, you can get home insurance quotes in 15 minutes or less — sometimes as quickly as 60 seconds — and it’s all thanks to artificial intelligence. “AI is able to digest and analyze vast datasets at unprecedented speeds and accuracy,” Ganje said.
For instance, a silicon valley startup called Cape Analytics harvests satellite image data, runs it through an AI network, then provides insurance companies with detailed information about homes and properties. “You can’t sit there and have humans reviewing pictures of 70 million homes in the event that somebody wants to get an insurance quote,” Cape Analytics co-founder and CEO Ryan Kottenstette told Fast Company last year.
Hippo, the insurance startup that boasts a “60-second quote” process, uses AI to make near-instantaneous estimates. “AI has allowed us to more accurately pre-fill data related to a home,” says Mike Gulla, Hippo’s senior director of underwriting. “This eliminates the need for customers to try and guess at the characteristics of their property, which can leave them underinsured in many instances.”
AI discrimination concerns
The customer-facing benefits of AI are certainly worth celebrating, but some experts are concerned about the introduction of what the analytical consulting firm McKinsey & Company calls “personalized pricing,” because it could cause AI to unintentionally discriminate against certain kinds of homeowners.
Two years ago, a study by ProPublica and Consumer Reports found residents of minority neighborhoods in California, Illinois, Missouri, and Texas paid higher auto insurance premiums — as much as 30% higher — than white neighborhoods with identical levels of risk. Insurers were also significantly less likely to fulfill claims in those minority neighborhoods. This discrepancy between communities is essentially the insurance equivalent of “redlining,” a term used to describe the denial or price-gouging of services in minority neighborhoods dating back to the 19th century.
James Lynch, chief actuary of the Insurance Information Institute, responded to the study with assurances that “[i]nsurance companies do not collect any information regarding the race or ethnicity of the people they sell policies to.” But now that home insurance rates and claims are increasingly processed with the help of artificial intelligence, a new factor could contribute to unintentional biases — an AI quirk called “proxy discrimination.”
Proxy discrimination happens when a system designed to be neutral nevertheless develops a bias against a particular subset. “Historically, [discrimination] occurred when a firm intentionally sought to discriminate against members of a protected class by relying on a proxy for class membership, such as ZIP code,” Anya Prince and Daniel Schwarcz wrote in the Iowa Law Review a few weeks ago. “However, proxy discrimination need not be intentional,” they say.
In other words, home insurance AI could develop a bias against particular communities without the insurance company even realizing it. For example, one of the main factors home insurance companies use to determine your premium is your credit history. But a study by the Urban Institute found “the difference in median credit scores is nearly 80 points” between minority communities and white communities. And that’s just one factor among dozens — how does an AI that calculates ZIP code data account for the effects of segregation, for instance? Prince and Schwarcz go on to claim “AI and big data are game changers when it comes to this risk of unintentional, but ‘rational,’ proxy discrimination.”
Whatever the costs and benefits of AI in home insurance, one thing’s for sure — we’ve only seen the tip of the iceberg. Tech startups like Lemonade are getting most of the headlines, but national providers like Allstate and The Hartford are already adopting AI, both internally and through third-party partners like Cape Analytics. And providers like Travelers and Hippo are already encouraging homeowners to install data-reporting devices. Ten years from now, or maybe even sooner, your home might be smart enough to submit an insurance claim for itself.